[Feat] (DirectVulkan, MG_Util): honour a glVertexAttribDivisor other than 1

Vulkan's VK_VERTEX_INPUT_RATE_INSTANCE advances an attribute once per instance and has
no way to say anything else, so every non-zero divisor collapsed to 1: an attribute the
application asked to change every three instances changed every one, and
KHR-GL40.draw_indirect.basic-drawArrays-instancing and its elements sibling drew the
wrong colours from instance one onward.

VK_EXT_vertex_attribute_divisor is exactly this state, so it is enabled when the device
has it and the per-binding divisors ride into the pipeline through
VkPipelineVertexInputDivisorStateCreateInfoEXT. Only divisors other than 1 are listed -
1 is what the plain input rate already means - and they join the layout hash, so two
layouts that differ only in a divisor no longer share a pipeline.

POST reports the feature either way, because without it the failure is silent and looks
like a shader bug: the attribute is fetched, just from the wrong instance. The GLES side
gains the two checks this session's other work made load-bearing for the same reason -
glPatchParameteri (without it GL_PATCH_VERTICES stays at the driver's 3 and a patch draw
of any other size renders nothing) and the transform feedback object entry points
(without them a second object cannot open a capture while the first is paused).

KHR-GL40.draw_indirect on Magma: 70/70 but for the arbitrary primitive-restart index,
which Vulkan cannot express at all.
This commit is contained in:
BZLZHH
2026-08-04 19:25:32 -04:00
parent a50b2c422b
commit cff959b2e8
5 changed files with 99 additions and 0 deletions
@@ -9894,6 +9894,36 @@ void main() {
MGLOG_W("VK_EXT_transform_feedback is unavailable; transform feedback capture will not work");
}
// VK_EXT_vertex_attribute_divisor. Vulkan's instance input rate advances an attribute
// once per instance and nothing else, so without this every glVertexAttribDivisor value
// collapses to 1 and an attribute meant to change every N instances changes every one.
m_vertexAttributeDivisorEnabled = false;
VkPhysicalDeviceVertexAttributeDivisorFeaturesEXT vertexAttributeDivisorFeatures{};
vertexAttributeDivisorFeatures.sType =
VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VERTEX_ATTRIBUTE_DIVISOR_FEATURES_EXT;
if (IsExtensionSupported(availableExtensions, VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME) &&
getPhysicalDeviceFeatures2 != nullptr) {
VkPhysicalDeviceFeatures2 featureQuery{};
featureQuery.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FEATURES_2;
featureQuery.pNext = &vertexAttributeDivisorFeatures;
getPhysicalDeviceFeatures2(m_physicalDevice.handle, &featureQuery);
if (vertexAttributeDivisorFeatures.vertexAttributeInstanceRateDivisor == VK_TRUE) {
if (!IsExtensionAlreadyEnabled(enabledDeviceExtensions,
VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME)) {
enabledDeviceExtensions.push_back(VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME);
}
vertexAttributeDivisorFeatures.vertexAttributeInstanceRateZeroDivisor = VK_FALSE;
vertexAttributeDivisorFeatures.pNext = const_cast<void*>(deviceCreateInfo.pNext);
deviceCreateInfo.pNext = &vertexAttributeDivisorFeatures;
m_vertexAttributeDivisorEnabled = true;
MGLOG_I("Enabled optional device extension: %s", VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME);
}
}
if (!m_vertexAttributeDivisorEnabled) {
MGLOG_W("VK_EXT_vertex_attribute_divisor is unavailable; a glVertexAttribDivisor other "
"than 1 will advance its attribute once per instance");
}
// Host query reset lets the occlusion-query ring recycle slots without a
// command-buffer round trip.
m_hostQueryResetEnabled = false;